Utilizing Machine Learning and Modeling in Predicting Functional Properties of Plant Proteins
Transforming future food design through data integration, predictive modeling, and computational structure-function mapping.
Date: November 17th at 10am CT
Platform: Zoom
Cost: FREE
Webinar Theme Overview
As the global demand for sustainable, plant-based foods accelerates, decoding the complex structure-function relationships of plant proteins has become pivotal to modern food innovation. This webinar highlights how cutting-edge machine learning (ML), computational modeling, and standardized data infrastructure are revolutionizing ingredient design and formulation.
Featuring leading researchers in food science and predictive analytics, the session bridges the gap between raw experimental data and actionable predictive modeling. Dr. Jing Zhao will introduce FoodProt, a USDA-NIFA-funded platform designed to standardize complex, literature-derived datasets and create analysis-ready repositories for reliable predictions. Complementing this data-centric foundation, Dr. Ronit Mandal will demonstrate how advanced algorithms, such as Gaussian Support Vector Regression and causal ML, can map macromolecular traits to critical functional behaviors like solubility, gelation, and emulsification.
Together, these presentations emphasize a transformative shift in food science: moving away from slow, trial-and-error laboratory experimentation toward precise, data-driven formulation. Attendees will discover how integrated database architectures and machine learning tools serve as indispensable drivers for accelerating sustainable protein innovation and rational food design.
Webinar Guests
From Heterogeneous Evidence to Trustworthy Prediction: FoodProt for Plant Protein Functionality
Dr. Jing Zhao
Associate Professor
School of Exercise and Nutritional Sciences
San Diego State University
Dr. Jing Zhao is an Associate Professor in the School of Exercise and Nutritional Sciences at San Diego State University. Dr. Zhao’ research focuses on the development, characterization (structure, functionalities, and flavor), and utilization of sustainable food proteins, such as plant, algae, and edible insect proteins. Dr. Zhao has served on the editorial board of the Journal of Food Science since 2019 and served as the division chair in 2024-2025. Dr. Zhao currently guest-edits a special issue on “Analytical Methods for Food Protein Characterization” for Sustainable Food Proteins and leads the USDA-NIFA funded FoodProt project.
Plant protein functionality data are abundant but difficult to compare because studies differ in protein source, processing history, analytical methods, units, measurement basis, and reporting completeness. This presentation introduces a USDA-NIFA funded project to build a high-quality protein functionality database – FoodProt. FoodProt is a web-based, provenance-aware platform developed to compile literature-derived and collaborative research data into reviewable, analysis-ready records. The platform will also provide basic data-visualization and machine-learning tools to support data analysis and prediction. The presentation will discuss the current development of FoodProt, key challenges in integrating heterogeneous datasets, recommended practices for data collection and reporting, and preliminary models used to evaluate the datasets and test the platform’s analytical capabilities.
Decoding Plant Protein Functionality: How Machine Learning Models are Revolutionizing Future Food Design
Dr. Ronit Mandal
Assistant Professor (Food Process Engineering)
Agricultural and Food Engineering Department
Indian Institute of Technology Kharagpur, Kharagpur
Dr. Ronit Mandal is an Assistant Professor of Food Process Engineering at the Indian Institute of Technology Kharagpur, specializing in AI applications in food processing, computational modeling, and dairy processing. He previously served as a postdoctoral researcher at the Plant Protein Innovation Center, University of Minnesota, focusing on computational modeling of plant protein functionality, following his PhD at the University of British Columbia. He has over forty publications and book chapters, and has received numerous prestigious accolades, including the IFT Nonthermal Processing Division-EFFoST Award, Indian Dairy Association Gold Medal, and UBC LFS Top Dissertation Award.
As specialized protein-based functional ingredients emerge for gelling, thickening, and emulsification applications, understanding the complex structure-function relationships of plant proteins is critical. Herein, we will discuss a comprehensive study evaluating 13 machine learning algorithms to predict key plant protein functional properties such as solubility, emulsifying activity, capacity, and gel strength using structural predictors like surface hydrophobicity, zeta potential, undenatured protein content, water holding capacity, soluble protein polymer content, ß-sheet content. The 𝐺𝑎𝑢𝑠𝑠𝑖𝑎𝑛 kernel-based Support Vector Regression model accurately predicted solubility, emulsifying activity index, emulsifying capacity, and gel strength of proteins using only a few macromolecular structural characteristics. Furthermore, we explored the application of causal machine learning to determine the true strength of causal relations between structural characteristics and functional properties of proteins subjected to various functionalization treatments. Therefore, we demonstrate how these computational models serve as indispensable tools for selecting optimal protein ingredients in modern food applications.
Webinar Moderator
